Validity and reproducibility of a web-based, self-administered food frequency questionnaire.

BACKGROUND/
OBJECTIVES: To assess the validity and the reproducibility of a newly developed web-based, self-administered food frequency questionnaire (web-FFQ). SUBJECTS/
METHODS: A total of 74 healthy subjects (34 men and 40 women) from the Québec City metropolitan area were asked to complete, in random order, the web-FFQ, a validated interviewer-administered FFQ (IA-FFQ) and a 3-day food record (3-day FR).
RESULTS: Mean intakes of 17/22 nutrients assessed between the web-FFQ and the 3-day FR were not significantly different (differences <10%, P≥0.11). Sex and energy-adjusted de-attenuated Pearson correlation coefficients for each nutrient varied from 0.12-0.98 (mean R=0.55, 95% confidence interval 0.46; 0.63) between the web-FFQ and the 3-day FR. All correlations were significant (P≤0.01) and above 0.34 (mean R=0.59, 95% confidence interval 0.54; 0.65) between the web-FFQ and the IA-FFQ, except for sodium (R=0.17, P=0.14). Cross-classification analysis revealed that on average, 77% of subjects were classified in the same or adjacent quartile of nutrient intake between the web-FFQ and the 3-day FR. Correlation coefficients for reproducibility of the web-FFQ tested 4-6 weeks apart in the same individuals were all equal or above 0.48 (P≤0.0001; mean R=0.72, 95% confidence interval 0.68; 0.76). More than 90% of the subjects were classified in the same or adjacent quartile between the two administrations of the web-FFQ, while only 0.8% was misclassified.
CONCLUSIONS: These data demonstrate that the newly developed web-based FFQ appears to have reasonable validity and good reproducibility for assessing nutrient intakes at the group and individual levels in a population of healthy adults.